SimToolReal: An Object-Centric Policy for Zero-Shot Dexterous Tool Manipulation

TL;DR

SimToolReal introduces a generalizable sim-to-real reinforcement learning approach for dexterous tool manipulation, capable of zero-shot generalization across diverse objects and tasks without task-specific training.

Contribution

It proposes a procedural generation method for diverse tool-like objects and trains a universal policy, reducing engineering effort and enabling zero-shot transfer in tool manipulation tasks.

Findings

Outperforms prior retargeting and fixed-grasp methods by 37%.

Matches performance of specialized RL policies on specific tasks.

Achieves strong zero-shot performance over 120 real-world rollouts.

Abstract

The ability to manipulate tools significantly expands the set of tasks a robot can perform. Yet, tool manipulation represents a challenging class of dexterity, requiring grasping thin objects, in-hand object rotations, and forceful interactions. Since collecting teleoperation data for these behaviors is challenging, sim-to-real reinforcement learning (RL) is a promising alternative. However, prior approaches typically require substantial engineering effort to model objects and tune reward functions for each task. In this work, we propose SimToolReal, taking a step towards generalizing sim-to-real RL policies for tool manipulation. Instead of focusing on a single object and task, we procedurally generate a large variety of tool-like object primitives in simulation and train a single RL policy with the universal goal of manipulating each object to random goal poses. This approach enables SimToolReal to perform general dexterous tool manipulation at test-time without any object or task-specific training. We demonstrate that SimToolReal outperforms prior retargeting and fixed-grasp methods by 37% while matching the performance of specialist RL policies trained on specific target objects and tasks. Finally, we show that SimToolReal generalizes across a diverse set of everyday tools, achieving strong zero-shot performance over 120 real-world rollouts spanning 24 tasks, 12 object instances, and 6 tool categories.